Pattern matrix /White Paper/Reasoning

ADPS Agent Design Pattern White Paper · Module Overview

Reasoning · Compiling Evidence into Reviewable Decisions

Reasoning artifacts, effort routing, parallel exploration, hypothesis testing, live interaction, and engineering acceptance.

Reasoning sits between perception and memory on one side and action on the other. Perception supplies current signals. Memory retrieves prior facts and experience. Reasoning turns them into a decision; action determines how that decision may change an external system.

A production system has to manage more than the final answer. A decision also needs evidence references, alternatives, uncertainty, budget use, a verification method, and a next step. Downstream components can then review, reject, or execute it under control.

A reasoning request is routed into serial, parallel, iterative, or layered reasoning before producing a structured decision
R2 selects resources and a reasoning path. R1, R3, R4, and R5 provide different decision structures. They may be composed; this is not a fixed pipeline.

The output boundary

“Approve the request” is only a conclusion. A decision that can cross a system boundary must state what supports it, when it remains valid, what is unresolved, and who may authorize the next action.

decision_id: dec_01K...
goal_ref: goal://incident/482
evidence_refs:
  - log://gateway/482#timeout
  - change://config/917
choice: rollback_recent_configuration
alternatives:
  - keep_observing
  - isolate_single_instance
uncertainty:
  level: medium
  unresolved: database latency has not been excluded
validation:
  before_action: reproduce on canary
  after_action: error rate returns to baseline
authority:
  required: on_call_approval
next_action: prepare_rollback_intent

This record is still a decision, not an execution command. The Action module must recheck authority, tools, arguments, current state, and acceptance conditions.

Five patterns, five engineering problems

PatternScopePrimary artifact
R1 Chain of ThoughtForms a sequential judgment and manages API-visible summaries, evidence, and model metadataReasoning summary, evidence binding, decision record
R2 Complexity-Based RoutingSelects a model, effort level, and fallback path from difficulty, risk, evidence gaps, and service objectivesRouteDecision, budget, fallback policy
R3 Parallel ExplorationRuns isolated candidate paths and aggregates them according to the cost of errorBranch results, disagreement record, aggregate decision
R4 Iterative Hypothesis TestingUpdates hypotheses with new evidence until convergence, budget exhaustion, or human escalationHypothesis tree, counter-evidence, exit reason
R5 Talker-ReasonerSeparates low-latency interaction from high-cost analysis and hands off structured stateTask packet, shared state, reasoning result packet

Difficulty and risk are separate decisions

A difficult problem does not automatically justify broader authority. A short request may still trigger a high-risk action. The router should score reasoning difficulty and business risk separately, then set the model tier, number of branches, time budget, and human boundary.

Low riskHigh risk
Low difficultyRules or a lightweight model; results are easy to recheckComputation may be simple, while evidence and approval stay strict
High difficultyDeep reasoning or parallel exploration under cost and latency limitsDeep reasoning, independent verification, explicit human review, and stop rules

Patterns can be nested

A production-incident workflow may use R2 to classify difficulty and risk. A common low-risk case enters R1. Conflicting evidence starts R3 so that independent branches can examine the incident. If the cause remains unclear, R4 advances through hypothesis, evidence collection, and falsification. While the user waits, the R5 Talker reports progress and the Reasoner continues in the background.

The composition needs an explicit convergence point. It waits for required branches, handles timeouts and conflicts, applies exit conditions, and reduces the result to one decision schema. More branches without convergence merely produce more answers.

Budget, exit, and escalation

ControlEngineering question
BudgetHow many tokens, model calls, concurrent branches, seconds, and external lookups are allowed?
EvidenceWhich claims require external facts, and are those facts still current?
DisagreementWhen do majority vote, any-alarm, and an independent judge match the cost of error?
ExitHow does the run stop on sufficient evidence, indistinguishable candidates, budget exhaustion, or a changed user goal?
EscalationWhich hypotheses, evidence, excluded paths, and open questions accompany a human handoff?

Interfaces with adjacent modules

ModuleInput to ReasoningOutput from Reasoning
PerceptionSignals, provenance, time, and parsed observationsRequests for another observation or clarification
MemoryVersioned, scoped facts, experience, and progressPublishable decision summaries and applicability boundaries
ActionTool results, business receipts, and current stateStructured decisions without inherited execution authority
ReflectionEvaluation results, failure attribution, and delayed outcomesDecision records and reproducible acceptance conditions
GovernanceAuthority, budget, prohibited operations, and human-review rulesRisk statements, pending intents, and evidence

Validation

  • Decision quality: accuracy, false positives, false negatives, and calibration on a business evaluation set.
  • Routing quality: misrouting and unnecessary escalation against the least costly acceptable path.
  • Convergence quality: useful falsification, limit exits, human escalation, and repeated evidence gathering.
  • Operating cost: quality, latency, tokens, concurrency, and external-tool cost reported together.
  • Reviewability: whether critical conclusions resolve to evidence, versions, rules, and accountable principals.

Published patterns

Suggested citation: ADPS, Reasoning: Compiling Evidence into Reviewable Decisions, Agent Design Pattern White Paper, 2026.

CC BY 4.0

Scope: This page defines module boundaries and engineering checks. It does not certify a model, product, or enterprise implementation. Scenarios illustrate pattern composition; named cases follow the evidence notes in their engineering case reports.

Chronicle

Recorded source
First Reasoning Module Workshop (26 August 2026); published ADPS Reasoning pattern specifications
Source date
First published on ADPS

View in the ADPS Chronicle